2 new functions

NAMESPACE

@@ -8,6 +8,7 @@ export(batch_pubmed_download2)
 export(check_pubmed_api_token)
 export(clean_journals_continents)
 export(convert_hex_to_char)
+export(detect_missing_journals)
 export(dygraph_year)
 export(extract_split_address)
 export(get_affiliation)
@@ -26,6 +27,7 @@ export(table_continent_year)
 export(table_country)
 export(table_country_journal)
 export(table_country_year)
+export(table_journal_count)
 export(table_missing_country)
 export(waffle_continent)
 export(waffle_continent_journal)

R/clean_journals_continents.R

@@ -7,13 +7,16 @@ clean_journals_continents <- function(data) {
       field = pubmedDashboard::journal_field$field[match(
         .data$journal, pubmedDashboard::journal_field$journal
       )],
+      field = ifelse(is.na(.data$field), pubmedDashboard::journal_field$field[match(
+        .data$journal, pubmedDashboard::journal_field$journal_short)], .data$field),
       original_journal = .data$journal %in% pubmedDashboard::journal_field$journal[1:6],
       continent = factor(.data$continent, levels = continent_order()),
       journal = clean_journal_names(.data$journal)
-      # journal = gsub(":.*", "", .data$journal),
-      # journal = tools::toTitleCase(.data$journal),
-      # journal = trimws(.data$journal),
-    )
+    ) %>%
+    dplyr::group_by(.data$journal) %>%
+    dplyr::mutate(first_Year = min(.data$year), last_year = max(.data$year),
+           year_range = paste0(.data$first_Year, "-", .data$last_year)) %>%
+    dplyr::ungroup()
 }
 
 #' @noRd
@@ -28,9 +31,42 @@ continent_order <- function(short = FALSE) {
 
 #' @noRd
 clean_journal_names <- function(journal) {
-  x <- gsub("&amp;", "&", journal, fixed = TRUE)
-  x <- gsub(":.*", "", journal)
+  x <- gsub("&amp;", "and", journal, fixed = TRUE)
+  # x <- gsub(" of the United States of America", "", x, fixed = TRUE)
+  x <- gsub(":.*", "", x)
   x <- gsub("[(].*", "", x)
   x <- tools::toTitleCase(x)
   trimws(x)
 }
+
+#' @title Detect missing journals
+#' @param data The processed dataframe of data
+#' @export
+detect_missing_journals <- function(data) {
+  data.frame(journal = pubmedDashboard::journal_field$journal_short) %>%
+    dplyr::mutate(found = pubmedDashboard::journal_field$journal_short %in%
+             clean_journal_names(unique(data$journal))) %>%
+    dplyr::arrange(data$found)
+}
+
+#' @title Count number of papers per journal, with year range
+#' @param data The processed dataframe of data
+#' @param datatable Whether to output a [DT::datatable] HTML table widget
+#'  instead of a regular dataframe (defaults to TRUE).
+#' @export
+table_journal_count <- function(data, datatable = TRUE) {
+  x <- dplyr::count(data, data$journal, data$field, data$year_range, sort = TRUE) %>%
+    dplyr::mutate(field = stringr::str_to_title(.data$field)) %>%
+    as.data.frame() %>%
+    dplyr::rename("year range" = "year_range") %>%
+    dplyr::rename_with(stringr::str_to_title)
+  if (isTRUE(datatable)) {
+    insight::check_if_installed("DT")
+    x <- DT::datatable(
+      x,
+      options = list(searching = TRUE, paging = TRUE),
+      caption = "Count of journals, with year range")
+  }
+  x
+}
+

data-raw/journal_field.R

@@ -64,7 +64,7 @@ general <- c(
   "Proceedings of the National Academy of Sciences of the United States of America",
   "Science (New York, N.Y.)",
   "Nature",
-  "Plos one"
+  "PloS one"
 )
 
 journal_field <- data.frame(